Aller au contenu. | Aller à la navigation

Outils personnels

Navigation

UMR 5672

logo de l'ENS de Lyon
logo du CNRS
Vous êtes ici : Accueil / Séminaires / Machine Learning and Signal Processing / Pauline Mouchès

Pauline Mouchès

CNNs and graph convolution networks for event detection in brain activity recordings of epilepsy patients
Quand ? Le 14/11/2023,
de 14:00 à 15:00
Où ? MGN1 105
Participants Pauline Mouchès
Ajouter un événement au calendrier vCal
iCal

ATTENTION : the time is 2pm exceptionnally,and in room MGN1 105  (previously R116) (ENSL, Site Monod, côté accueil, 1er étage).

Speaker: Pauline Mouchès (CRNL)

https://scholar.google.fr/citations?user=KK6PANoAAAAJ&hl=fr

Title: CNNs and graph convolution networks for event detection in brain activity recordings of epilepsy patients.

Abstract: Brain activity recordings of epilepsy patients contain brief morphologically defined events between seizures, called spikes. Their detection allows the localization of the brain region from which seizure originate. Nevertheless, they are often detected manually by clinicians, which is a time-consuming task. In this presentation, I will introduce our recent work consisting in coupling a CNN with a Graph convolutional network for automated spike. The task is addressed as a classification problem using short time windows from the recordings represented as graph data. I will discuss the challenges encountered with the data including their imbalance nature and imperfect labels, and propose an approach to integrate clinicians' feedback in the model training loop.

arXiv preprint arXiv:2310.09236